Calcium(2+)-dependent annexin self-association on membrane

Phosphorylation Mutants Elucidate the Mechanism of Annexin IV-Mediated .... Swathi Krishnan , Kaitlin E. Swanson , Judy U. Earley , Elizabeth M. McNal...
0 downloads 0 Views 2MB Size
Biochemistry 1991, 30, 9607-961 5 teria, Chapter 11, pp 235-254, Ellis-Harwood Ltd., ChiChester, England. Waygood, E. B. (1986) Biochemistry 25, 4085-4090. Waygood, E. B., & Steeves, T., (1980) Can.J . Biochem. 58, 40-48. Waygood, E. B., Meadow, N. D., & Roseman, S . (1979) Anal. Biochem. 95, 293-304. Waygood, E. B., Mattoo, R. L., & Peri, K. G. (1984) J . Cell Biochem. 25, 139-1 59. Waygood, E. B., Erickson, E., El-Kabbani, 0. A. L., & Delbaere, L. T. J . (1985) Biochemistry 24, 6938-6945. Waygood, E. B., Reiche, B., Hengstenberg, W., & Lee, J. S.

9607

(1987) J . Bacteriol. 169, 2810-2818. Waygood, E. B., Pasloske, K., Delbaere, L. T. J., Deutscher, J., & Hengstenberg, W. (1988) Biochem. Cell Biol. 66, 76-80. Waygood, E. B., Sharma, S., Bhanot, P., El-Kabbani, 0. A. L., Delbaere, L. T. J., Georges, F., Wittekind, M. G., & Klevit, R. E. (1989) FEMS Microbiol. Rev. 63, 43-52. Wittekind, M., Reizer, J., & Klevit, R. E. (1990) Biochemistry 29, 7191-7200. Yanisch-Perron, C., Vieira, J., & Messing, J. (1985) Gene 33, 103-1 19. Zoller, M. J., & Smith, M. (1984) DNA 3, 479-488.

Ca*+-DependentAnnexin Self-Association on Membrane Surfaces? William J. Zaks* and Carl E. Creutz* Department of Pharmacology, University of Virginia, Charlottesville, Virginia 22908 Received June 14, 1991

ABSTRACT: Annexin self-association was studied with 90' light scattering and resonance energy transfer

between fluorescein (donor) and eosin (acceptor) labeled proteins. Synexin (annexin VII), p32 (annexin IV), and p67 (annexin VI) self-associated in a Ca2+-dependent manner in solution. However, this activity was quite labile and, especially for p32 and p67, was not consistently observed. When bound to chromaffin granule membranes, the three proteins consistently self-associated and did so at Ca2+levels (pCa 5.0-4.5) approximately 10-fold lower than required when in solution. Phospholipid vesicles containing phosphatidylserine and phosphatidylethanolamine (1: 1 or 1:3) were less effective at supporting annexin polymerization than were those containing phosphatidylserine and phosphatidylcholine (1:0, 1:1, or 1:3). The annexins bound chromaffin granule membranes in a positively cooperative manner under conditions where annexin self-association was observed, and both phenomena were inhibited by trifluoperazine. Ca2+-dependent chromaffin granule membrane aggregation, induced by p32 or synexin, was associated with intermembrane annexin polymerization a t Ca2+levels less than pCa 4, but not a t higher Ca2+ concentrations, suggesting that annexin self-association may be necessary for membrane contact at low Ca2+ levels but not at higher Ca2+ levels where the protein may bind two membranes as a monomer.

T e annexins are a newly described group of homologous proteins that bind phospholipid membranes in a Ca2+-dependent manner [for reviews see Klee (1988), and Burgoyne and Geisow (1989)l. Some members of this group are also commonly known as lipocortins (Huang et al., 1986), calpactins (Glenney, 1986), chromobindins (Creutz et al., 1983, 1987), calelectrins (Sudhof et al., 1984), or placental anticoagulant proteins (Tait et al., 1988). Currently, 10 distinct members of this gene family, termed annexins I-X, have been identified (Pepinsky et al., 1988; Burns et al., 1989; Hauptmann et al., 1989; Johnston et al., 1990). Comparison of their amino acid sequences reveals a common structural theme: Each protein has two regions, a variable-length amino-terminal region lacking homology with other members of the family and a core region of four or eight repeating 70 amino acid domains, which share 40-6096 homology between family members. The biological function of the annexin proteins is unknown, but they have been hypothesized to play a role in signal transduction (Hollenberg et al., 1988), in exocytosis (Creutz +This study was supported by a grant from the NIH (DK33151). C.E.C. was supported by an Established Investigator award from the American Heart Association with funds contributed in part by the Virginia Affiliate. * Correspondence may be addressed to either author.

0006-2960/91/0430-9607$02.50/0

et al., 1978, 1983), in the organization of membrane phospholipid domains (Geisow et al., 1987), as structural/regulatory elements of the cytoskeleton (Glenney, 1986), and as regulators of phospholipase A2 (Huang et al., 1986). These theories are based on the preferential cellular localization of some of these proteins to the plasma membrane/cortical cytoskeleton and their ability to bind in a Ca2+-dependent manner to phospholipid membranes and in some cases to cytoskeletal elements (Geisow et al., 1987). However, the mode of interaction of the annexins with phospholipids and proteins at the membrane surface is unclear. In addition to interactions with other proteins, some annexins have been reported to self-associate. This phenemonon was first reported for isolated synexin in solution, which formed 50 X 150 %, rods, bundles of rods, and paracrystalline arrays in a Caz+-dependentmanner (Creutz et a]., 1979). A similar self-association event was seen with isolated Torpedo calelectrin, which formed morphologically different structures: circular forms composed of 50-A globular subunits in the absence of calcium and amorphous aggregates of polygonal structures each composed of the 50-A globular subunit in the presence of calcium (Walker et al., 1983). Other reports, however, have claimed that other members of the annexin family such as p32 (annexin IV) or calpactin (annexin 11) do not self-associate (Shadle et al., 1985). Similarly, it has been debated whether annexins self-associate on membrane surfaces. Some investigators have 0 1991 American Chemical Society

9608 Biochemistry, Vol. 30, No. 40, 1991 reported cooperative annexin-membrane binding interactions (Zaks & Creutz, 1990) or visualization of membrane-bound annexin aggregates (Newman et al., 1989, 1991; Mosser et al., 1991), while others report only noncooperative membrane binding phenomena (Tait et al., 1989; Meers, 1990). In this paper we show that a resonance energy transfer assay may be used to monitor the self-association of three annexin proteins: synexin (annexin VII; Creutz et al., 1978), p32 (annexin IV; Shadle et al., 1985; Geisow et al., 1986), and p67 (annexin VI; Owens & Crumpton, 1984; Shadle et al., 1985; Sudhof et al., 1984). Each of the three proteins was shown to be capable of Ca2+-dependent self-association in solution and on some membrane surfaces. Furthermore, the self-association event enabled annexins to bind membranes in a positively cooperative manner and appeared to play a role in the ability of the annexins to aggregate membranes at Ca2+ levels of less than 100 pM. EXPERIMENTAL PROCEDURES Materials. Fluorescein 5-isothiocyanate and eosin %sothiocyanate were purchased from Molecular Probes, Inc. Disuccinimidylsuberate was obtained from Pierce. Trifluoroperazine (TFP),' obtained from Smith, Kline and French, was a gift from Dr. M. J. Peach (University of Virginia). Phospholipids were from Sigma. Sephadex G-25 (PD-IO) columns were obtained from Pharmacia Fine Chemicals. Preparation of Chromagin Granule Membranes and Phospholipid Vesicles. Chromaffin granule membranes were prepared by the method of Bartlett and Smith (1974). Multilamellar liposomes were prepared as follows. Chloroform solutions of phospholipids were mixed in various proportions, dried under nitrogen, and resuspended at 2 mg/mL by vortex mixing in buffer containing 0.24 M sucrose, 30 mM KCl, and 40 mM Hepes, pH 7.0. Preparation of Annexins and Fluorescent Derivatives. Synexin was prepared from bovine liver as previously described (Zaks & Creutz, 1990). The protein was routinely reprecipitated in 20% saturated ammonium sulfate after the final FPLC chromatography step, to give a preparation greater than 85% pure as determined from Coomassie-stained SDS gels. The mammalian calelectrins (p32 and p67) were isolated from bovine liver as described (Creutz et al., 1987), and each protein was estimated to be greater than 95% homogeneous. For fluorescent labeling, 250 pg of synexin, p32, or p67 in buffer containing 0.24 M sucrose, 30 mM KCl, and 40 mM Hepes, pH 8.0, was allowed to react with a 10-fold molar excess of fluorescein 5-isothiocyanate (FITC) or eosin 5-isothiocyanate (EITC) (added as a concentrated solution in dimethylformamide) for 20 min at 4 OC. The unreacted free FITC or EITC was removed from the protein by Sephadex G-25 chromatography in the presence of a buffer containing 0.24 M sucrose, 30 mM KCl, and 40 mM Hepes, pH 7.0. The concentration of bound dye was determined by the absorbance with a molar extinction coefficient of 42 500 M-' cm-' at 495 nm for FITC and 84000 M-' cm-' at 524 nm for EITC. In the various experiments the concentration of bound dye ranged from 0.2-0.4 mol of dye/mole of annexin. The labeled proteins were found to retain full chromaffin granule aggregating activity. Self-Association Assays. The 90° light scattering from solutions of synexin, p32 or p67 was monitored at 350 nm in I Abbreviations: Hepes, N-(2-hydroxyethyl)piperazine-N'-2-ethanesulfonic acid; FITC, fluorescein 5-isothiocyanate; EITC,eosin s-isothiocyanate; HEDTA,N-(hydroxyethy1)ethylenediaminetriacetic acid; TFP,trifluoperazine.

Zaks and Creutz a 0.5 cm X 0.5 cm quartz cuvette as described (Sterner et al., 1985). Fluorescence experiments were performed on a SPEX Fluorolog 2 model 111C spectrofluorometer in a 0.5 cm X 0.5 cm quartz cuvette at 22 OC. Fluorescein- and eosin-labeled annexin proteins were mixed in a 1:l molar ratio in 0.24 M sucrose, 30 mM KCl, and 40 mM Hepes, pH 7.0, buffer in the absence or presence of chromaffin granule membranes. The samples were excited at 470 nm (slit width 1.25 or 2.5 mm) and fluorescence emission was recorded either at 510 nm (slit width 2.5 mm) as a function of time or a 519 and 534 nm (slit width 2.5 mm) after 3 min of annexin self-association when polymerization reach a plateau. Annexin self-association was initiated by adding Ca2+ in the form of Ca2+-HEDTA buffers prepared and standardized with a calcium electrode as previously described (Zaks& Creutz, 1990). Reported pCa values represent Ca2+activity and not Ca2+concentration. To compare these values with pCa values reported in concentration units, all values of pCa given in this paper should be decreased by approximately 0.26 units [Le., pCa 6.0 (activity) becomes pCa 5.74 (concentration)]. Membrane-Binding Assays. Binding of FITC-labeled or iodinated annexin proteins to chromaffin granule membranes was measured by a centrifugation assay as previously described (Zaks & Creutz, 1990). Binding was quantitated either as the amount of radioactivity in the twice-washed pellet or as the decrease in fluorescence (A, = 470, A,, = 519) intensity of the supernatant relative to an identical sample in the presence of 2.5 mM HEDTA. Both assays gave essentially identical results. Cross-Linking Experiments. p32 (50 pg/mL) was incubated for 10 min at 4 OC with a 5-, lo-, 50-, and 100-fold molar excess of disuccinimidylsuberate (added as a 100-fold concentrated solution in dimethylformamide) in 100 pL of buffer containing 0.24 M sucrose, 30 mM KCl, 40 mM H e w , pH 8.0, 100 pg/mL phospholipid (PS/PC 1:2, w/w), and 2.5 mM HEDTA with 0 or 3.5 mM CaC12. The reactions were quenched by adding 5 pL of a saturated solution of Tris, and after 5 min the pH was neutralized with HCl. The mixture of cross-linked proteins was separated on a 10% SDS-polyacrylamide gel (Laemmli, 1970) and visualized by staining with silver (Morrissey, 198 1). Miscellaneous Procedures. Annexin-induced aggregation of chromaffin granule membranes was measured by a turbidity (at 540 nm) assay as previously described (Creutz et al., 1978). Protein concentration was measured by the method of Bradford (1976) with bovine serum albumin as the standard for p32, p67, and chromaffin granule membranes and bovine gamma globulin as the standard for synexin. With these standards, this assay gives protein concentrations that agree well with those determined by the Lowry assay (Pollard et al., 1978). RESULTS Energy Transferas a Measure of Annexin Self-Association. The Ca2+-dependentself-associationof synexin in solution has been detected by 90° light-scattering measurements (Sterner et al., 1985; Creutz et al., 1979) and by electron microscopic visualization of negatively stained synexin polymers (Creutz et al., 1979). In order to determine whether this process could be measured by a resonance energy transfer assay, the Ca2+-dependent polymerization of mixtures of fluorescein (fluorescence donor) and eosin (fluorescence acceptor) labeled synexin was measured by both 90' light scattering and fluorescence donor (fluorescein) quenching at 5 10 nm. At this wavelength, representing the shoulder of the fluorescein emission spectrum, eosin makes only a negligible ( 5 % ) con-

Biochemistry, Vol. 30, No. 40, 1991 9609

Annexin Self-Association on Membrane Surfaces

I

y

I

I

40

I

1:

W

30

X V

m

9 8 z

z

W

lo

20

c

z

I-

2

e a W a

W V

5

log a

PCa

0

Ca2+dependence of synexin self-association in solution. The self-association of a 1:l mixture of fluorescein- and eosin-labeled synexin (1 -76pg/mL) was measured by donor (fluorescein)quenching at 5 10 nm ( 0 )or light scattering (m) as described under Experimental Procedures. FIGURE 1:

tribution to the fluorescence signal, and the efficiency of energy transfer can be determined directly from the reduction in fluorescein quantum yield. As shown in Figure 1, measurement of donor quenching detects an event with a similar CaZ+ dependence to that of synexin polymerization determined by light scattering. To determine whether this donor quenching was due to energy transfer, the magnitude of fluorescein quenching in a 1:1 mixture of fluorescein-labeled annexin and unlabeled annexin was compared with that seen with a 1:l mixture of fluorescein- and eosin-labeled annexin. It was observed that fluorescein quenching was enhanced 2.1-fold in the presence of eosin-labeled protein. Therefore, the quenching of the donor fluorescence was apparently primarily due to the interaction between the fluorescein- and eosin-labeled species. In studies conducted in the presence of membranes (see below) donor quenching was found to be similarly dependent upon the presence of the acceptor. Since fluorescein quenching was found to be insensitive to Ca*+/phospholipid binding (Figure 3B), the donor quenching seen in the absence of eosin-labeled protein appears to represent fluorescein self-quenching. As an alternate measure of energy transfer, we also examined the ratio of acceptor (eosin) to donor (fluorescein) fluorescence, measured at 543 and 5 19 nm, respectively. Coincident with the decrease in the fluorescence at 510 nm was an increase in the 543/5 19 ratio, consistent with transfer of energy from the fluorescein to the eosin. However, the change in this ratio was due primarily to the reduction in donor fluorescence since a Caz+-dependent increase in eosin (acceptor) fluorescence was not observed in this study, possibly due to self-quenching of eosin at high concentrations in the annexin aggregates. Annexin Self-Association in the Presence of Membranes. Although synexin self-associates in solution, there have been no direct studies to determine whether this self-association event occurs when the protein is membrane bound. However, we have previously shown that synexin binding to membranes shows positive cooperativity (Zaks & Creutz, 1990), which is suggestive of such a process. In order to provide more direct evidence for self-association on the membrane surface, we measured the Ca2+-dependentenergy transfer in a mixture of fluorescein- and eosin-labeled synexin in the presence of chromaffin granule membranes. As shown in Figure 2B, the extent of energy transfer was greatly enhanced in the presence of the membranes compared to that seen in the absence of membranes, as shown in Figure 1. These data suggest that synexin polymerizes on the membrane surface, possibly to a greater extent than in solution. However, we cannot eliminate the possibility that the greater magnitude of energy transfer

g